RESULTADO DE LA PROPUESTA A TRAVÉS DE UN MODELO DE COACHING EN AGROPRIM FARMS CIA LTDA.

A novel class of protein reagents designed to modulate tissue micromilieus was developed in this study. The work focused primarily on one example as proof of concept: CXCL10-mucin-GPI – a fusion protein consisting of a CXCL10 chemokine head, fused to the mucin domain taken from CX3CL1, and a C-terminal GPI anchor signal sequence substituting for the transmembrane domain of CX3CL1. Based on the biology of CXCL10, this reagent should allow the selective recruitment of cytotoxic CXCR3+ leukocytes into solid tumors. In parallel, a series of constructs were generated to act as controls: CXCL10-GPI, identical to CXCL10-mucin-GPI but lacking the mucin domain, should serve as a control for effects mediated by this domain. A membrane-bound, GPI-anchored EGFP protein was constructed to act as a GPI-anchored negative control protein lacking any obvious function in leukocyte recruitment. In addition, soluble versions of both CXCL10 fusion proteins (CXCL10-Stop and CXCL10-mucin-Stop) were designed as controls for the GPI anchor.

3.1.1

Cloning of the recombinant fusion genes

The fusion constructs were cloned as described in 2.2.3 and the respective compositions are summarized in Figure 9 A. Briefly, the GPI anchor signal sequence from LFA-3 had been previously cloned from a human inflamed kidney cDNA sample and inserted into a pUC19 plasmid (Notohamiprodjo et al. 2006). The DNA sequence encoding the double c-myc epitope tag that was used to facilitate purification and detection strategies was amplified by PCR from the MP71 gp100 vector and inserted into the pUC19 plasmid, 5´ of the GPI signal sequence. Subsequently, the coding sequence of human CXCL10 was ligated in-frame into the plasmid, 5´ of the double c-myc tag and the GPI anchor signal sequence. The resulting construct termed CXCL10-GPI was subcloned into a pEFdhfr vector for expression in dihydrofolate reductase deficient Chinese hamster ovary (CHOdhfr-/-) cells. The CXCL10 gene was then further subcloned into a pEFdhfr plasmid containing the mucin domain of CX3CL1, which had been amplified from a human inflamed kidney cDNA sample, a double c-myc tag and the GPI signal sequence from LFA-3. In the resulting construct termed CXCL10-mucin-GPI, the CXCL10 gene thus directly preceded the mucin domain sequence, followed by the c-myc tag and the GPI anchor signal sequence.

In addition to the genes for the surface anchored proteins, constructs were generated in which the GPI signal sequence was replaced by a stop codon. The resulting CXCL10-Stop and CXCL10-mucin- Stop proteins were expected to be secreted into the supernatant and should serve as non-anchored control proteins.

Finally, the sEGFP-GPI construct was created to act as a GPI-anchored control protein. It was generated by fusing the secretion signal sequence of human TIMP-1 N-terminally (5´) to the gene sequence of EGFP followed by a double c-myc tag and the GPI signal sequence from LFA-3. Figure 9 shows a schematic overview of the various recombinant genes (A) as well as an electrophoretic analysis of the respective plasmids (B). For this analysis, all pEF-plasmids were digested with EcoRI and SalI and the resulting fragments were separated in a 1% agarose gel.

As seen in the electrophoretic analysis, the CXCL10-GPI construct was 500 bp long, 317 of which constituted the CXCL10 chemokine domain at the 5´ end. The c-myc tag accounted for the following 66 bp, and the GPI anchor signal sequence for the last 117 bp. CXCL10-mucin-GPI had a total length of 1235 bp with the mucin domain, located between CXCL10 and the c-myc tag, accounting for the additional 735 bp (245 amino acids) compared to CXCL10-GPI and thus forming the biggest part of the resulting protein. The soluble constructs CXCL10-Stop and CXCL10-mucin-Stop were both 112 bp shorter than their GPI-anchored counterparts due to the lack of the GPI-anchor signal sequence resulting in total lengths of 388 and 1123 bp, respectively. The sEGFP-GPI gene sequence had a total length of 978 bp. The first 75 bp of the gene encoded the secretion signal sequence of TIMP-1, followed by the 720 bp long EGFP sequence and again 66 bp for the c-myc tag and 117 bp for the GPI anchor sequence. The exact DNA sequences of all constructs can be found in 5.7.

Figure 9: The recombinant CXCL10 and EGFP fusion genes display the anticipated lengths. A: Schematic overview of the different recombinant genes that were cloned in the present study. The scale indicates a length of 100 bp. Abbreviations: s = secretion signal sequence of TIMP-1; m = double c-myc epitope tag. B: Electrophoretic analysis of the various recombinant genes. 200 ng of pEF- CXCL10-mucin-GPI, pEF-CXCL10-mucin-Stop, pEF-CXCL10-GPI, pEF-sEGFP-GPI or 600 ng of pEF- CXCL10-Stop were digested with EcoRI and SalI to cut out the entire coding sequences and the fragments were separated in a 1% agarose gel. DNA bands were visualized under UV light using ethidium bromide staining. An inverted image of the resulting gel is shown to visualize the different sizes of the recombinant constructs.

3.1.2

Expression and detection of the recombinant fusion proteins

Chinese hamster ovary (CHO) cells were chosen for the expression of the recombinant proteins for two reasons: First, only eukaryotic cells are capable of attaching GPI anchors to proteins, which precluded the use of bacterial expression systems. Second, a mammalian cell line was chosen in order to keep the glycosylation pattern of the mucin domain similar to that found on human proteins. All plasmids were linearized prior to transfection and dihydrofolate reductase (dhfr)- deficient CHO cells that are auxotrophic for hypoxanthine and thymidine were transfected by electroporation as described in 2.2.2.14. The dhfr gene in the pEFdhfr plasmids enabled the selection of successfully transfected cells on the basis of its ability to complement the auxotrophy of CHOdhfr-/- cells. Dialyzed serum devoid of nucleotides was therefore used in the medium to generate selective pressure. The transfected CHO-CXCL10-GPI and CHO-CXCL10-mucin-GPI cells were subsequently additionally treated with methotrexate, an inhibitor of the dihydrofolate reductase, to eliminate cells with a low expression level as described in 2.2.2.14. By means of this treatment, cells with a low expression level of the transgenes could be eliminated, as evidenced by the loss of a moderately positive population in FACS staining (data not shown).

3.1.2.1 Subunits of the GPI-anchored fusion proteins can be detected on transfected CHO cells FACS staining was performed to evaluate the expression of the recombinant fusion proteins in transfected CHO cells. To this end, the cells were stained with antibodies specific for the c-myc epitope tag, the CX3CL1 mucin domain or CXCL10 and subsequently analyzed by flow cytometry as described in 2.2.5.1.1. Figure 10 shows the results of a representative FACS staining.

As shown in the figure, the GPI-anchored CXCL10 fusion proteins could readily be detected on the surface of the transfected cells. As expected, both CHO-CXCL10-GPI and CHO-CXCL10-mucin-GPI cells stained positively for the c-myc epitope tag and the CXCL10 chemokine head, while only CHO- CXCL10-mucin-GPI cells stained positively also for the mucin domain. The cells that had been transfected with the soluble CXCL10-mucin-Stop construct could not be stained with any of the antibodies, indicating that this fusion protein was (due to the lack of a GPI anchor) secreted into the medium, where it could also be detected by western blotting (data not shown). The same was found for CXCL10-Stop transfected cells, suggesting that fusion proteins lacking a GPI anchor did not bind nonspecifically to the CHO cells (data not shown). On CHO-sEGFP-GPI cells, the c-myc epitope tag could be detected as well as the fluorescence of the EGFP protein, indicating that the protein structure of EGFP was not compromised by the membrane-associated expression.

3.1.2.2 The GPI-anchored fusion proteins are targeted to the cell membranes of transfected cells The FACS analysis described above demonstrated surface expression of the recombinant fusion proteins. As an additional verification of correct targeting of the proteins to the cell membrane, immunofluorescence microscopy was performed as detailed in 2.2.5.3, because this approach is able to more precisely identify the subcellular localization of the detected proteins. To this end, the cells were grown in special dishes developed for fluorescence microscopy, fixed, and subsequently stained with c-myc specific antibodies. A combination of biotinylated secondary antibodies and RPE-labeled streptavidin was used to detect bound antibodies. Figure 11 shows representative images acquired on an inverted fluorescence microscope.

Figure 10: Subunits of the recombinant GPI-anchored proteins can be detected on the surface of transfected CHO cells. A: Stably transfected CHO cells were incubated with antibodies against the c-myc epitope tag, the mucin domain, the CXCL10 chemokine head or matching isotype controls. Bound antibodies were detected by staining with FITC-conjugated secondary antibodies and the fluorescence intensity was measured by FACS (FL-1H). Black lines indicate staining with the isotype controls, blue lines indicate staining with the specific antibodies. B: In the case of sEGFP-GPI transfected CHO cells, RPE-conjugated secondary antibodies were used to detect the c-myc epitope tag (anti c-myc antibodies: blue line, isotype control: black line) while the fluorescence by the EGFP- protein was measured in the FL-1 channel (FL1-H; green line) and compared to non-transfected CHO cells (grey line). All histograms are gated on viable cells identified by 7-AAD exclusion.

For all cells transfected with GPI-anchored fusion proteins, a strong staining by the c-myc specific antibodies was observed. The signal was most pronounced at the periphery of each cell, indicating that the fusion proteins were expressed in a membrane-associated manner. In contrast, the cells transfected with the soluble CXCL10-mucin-Stop construct did not stain positively. This finding was expected as the lack of a GPI anchor in CXCL10-mucin-Stop lead to secretion of the expressed protein into the medium. An absence of surface staining was also found for non-transfected cells, showing that the signal obtained with the anti c-myc antibodies was specific. In the case of sEGFP-GPI transfected cells, the fluorescence by EGFP was additionally detected at the cell surface. In summary, these results showed that the GPI-anchored proteins had successfully been targeted to the cell membrane by the addition of a GPI anchor signal sequence.

3.1.2.3 The N-termini of the CXCL10 fusion proteins are correctly processed

The N-terminus of chemokines is vitally important for their function. Changes in the amino acid composition of the N-terminus can lead to dramatic changes in the physiologic activity of the respective chemokine. For that reason, the N-terminal amino acid sequence of one of the CHO-cell- expressed CXCL10 fusion proteins was exemplarily determined using Edman sequencing. CXCL10- mucin-Stop was analyzed because for the soluble protein it was possible to obtain protein amounts sufficient for Edman sequencing from cell culture supernatants. The analyses were conducted in collaboration with R. Mentele from the Max-Planck Institute for Biochemistry in Martinsried. CXCL10- Figure 11: The recombinant GPI-anchored proteins are expressed in a membrane-associated manner. The indicated stably transfected CHO cell lines or non-transfected CHO cells (CHO) were grown in special fluorescence microscopy dishes, fixed and subsequently stained with anti c-myc primary and biotin-conjugated secondary antibodies followed by RPE-labeled streptavidin. In the case of CHO-sEGFP-GPI, the fluorescence of the EGFP protein was additionally detected using appropriate filters. All images within each horizontal row were acquired using the same exposure time. The bars in each image indicate a length of 50 µm.

mucin-Stop protein purified from cell culture supernatants as detailed in 2.2.4.1.4 was subjected to SDS-PAGE and subsequently transferred onto a PVDF membrane. As control, commercially available recombinant CXCL10 that had been demonstrated to be bioactive by the manufacturer and also in our own assays (see below) was identically treated. The membrane was stained and protein bands displaying the expected sizes were cut out and subjected to the sequencing procedure.

The results showed that both recombinant conventional CXCL10 and CXCL10-mucin-Stop comprised a mixture of proteins with either the sequence V-P-L or P-L-S at the N-terminus. The V-P-L sequence corresponds to the mature N-terminus of the well-documented, naturally occurring 77 amino acid form of CXCL10, while P-L-S would correspond to a 76 amino acid form. The latter form could have originated either from imprecise cleavage of the signal sequence during the secretion process, or from post-translational proteolytic modification which might also occur during the electrophoretic procedure. Because the mixture occurred in both protein samples, with the commercial protein having been demonstrated to be bioactive, the N-terminus of CXCL10-mucin-Stop was considered to be correctly processed by the CHO cells. As all CXCL10 fusion proteins were expressed in the same cell type from the same vector and the same CXCL10 gene sequence, it was assumed that the signal sequence was also correctly cleaved in the other fusion proteins. This consideration was complemented by several bioactivity experiments that were performed later on with the GPI- anchored proteins (see 3.2, 3.6 and 3.7).